CN106411605A - Node network self-organizing method, apparatus, server and system - Google Patents
Node network self-organizing method, apparatus, server and system Download PDFInfo
- Publication number
- CN106411605A CN106411605A CN201610963725.XA CN201610963725A CN106411605A CN 106411605 A CN106411605 A CN 106411605A CN 201610963725 A CN201610963725 A CN 201610963725A CN 106411605 A CN106411605 A CN 106411605A
- Authority
- CN
- China
- Prior art keywords
- node
- meshed network
- task
- port numbers
- self
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/12—Discovery or management of network topologies
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Information Transfer Between Computers (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a node network self-organizing method, apparatus, server and system. The method comprises the following steps: receiving task starting instructions sent by a master control node; selecting port numbers used for tasks; returning host names of slave nodes and the selected port numbers to the master control node so as to enable the master control node to generate a node network graph according to the tasks started by each slave node and the returned host names and the port numbers; and receiving the node network graph sent by the master control node. According to the technical scheme, the port numbers are distributed by each slave node rather than the master control node such that conflicts between the port numbers distributed by the master control node and the port numbers already used by the slave nodes are avoided, and at the same time, the generated node network graph can be used for management of a node network and connection establishment between the slave nodes such that the application demand is satisfied and the success rate of construction of the node network is also improved at the same time.
Description
Technical field
The present invention relates to technical field of the computer network is and in particular to a kind of meshed network self-organizing method, device, service
Device and system.
Background technology
There are multiple nodes in distributed type assemblies, these nodes can run the same task of executed in parallel.In a lot of situations
Under, executing tasks parallelly also needs to node and enabling communication between nodes, and the tissue of therefore meshed network is asking of non-the normally off key
Topic.In prior art, generally pass through main controlled node in task when setting up, be each execution task from node distribution execution this
The port numbers that the process of business is used, so each from node just can know that the process executing this task in other from nodes is made
Port numbers, thus set up with other from nodes be connected.But, the port numbers of distribution actually with a pair of each from node 1
Answer, main controlled node, without the port numbers knowing use on a certain node, is assigned with this node in assignment of port numbers
The port numbers having used, will affect the startup of task.For example, node 1 running of task A uses port 8080, is leading
Specify port 8080, then just cause port collision during the port numbers that control node is used by node 1 distribution task B again.
Content of the invention
In view of the above problems it is proposed that the present invention so as to provide one kind overcome the problems referred to above or at least in part solve on
State meshed network self-organizing method, device, server and the system of problem.
According to one aspect of the present invention, there is provided a kind of meshed network self-organizing method, including：
Receive the task start instruction that main controlled node sends；
Choose the port numbers for this task；
The port numbers of the host name of this from node and selection are returned to described main controlled node, so that described main controlled node root
Starting according to each from node of task and the host name of return and port numbers, generate meshed network figure；
Receive the meshed network figure that described main controlled node sends.
Alternatively, the described port numbers chosen for this task include：
From this from node currently unappropriated port numbers, randomly select a port number.
Alternatively, the described meshed network figure receiving described main controlled node transmission includes：
Periodically send, to described main controlled node, the request obtaining meshed network figure, receive described main controlled node according to this request
The meshed network figure returning；
And/or,
Receive the meshed network figure that described main controlled node active issues.
Alternatively, the method also includes：
According to described main controlled node send meshed network figure, with this meshed network in figure other one or more from section
Point is set up and is connected.
Alternatively, described task start instructs the enabled instruction for deep learning subtask；Described deep learning subtask
Including：Parameter server subtask and/or worker subtask.
According to another aspect of the present invention, there is provided a kind of meshed network self-organizing method, including：
According to the mission bit stream of input, send task start instruction to one or more from nodes；
Receive host name and the port numbers that each from node returns；
Being started according to each from node of task and the host name of return and port numbers, generate meshed network figure；
Described meshed network figure is sent to one or more from nodes.
Alternatively, described described meshed network figure is sent to one or more from nodes includes：
When receive from node transmission acquisition meshed network figure request when, by described meshed network figure be sent to this from
Node；
And/or,
Described meshed network figure is sent to all from nodes being connected with this main controlled node.
Alternatively, described mission bit stream is the mission bit stream of deep learning task；Described mission bit stream includes：For executing
The number of nodes of deep learning task, deep learning subtask type, all types of subtask quantity.
Alternatively, the described mission bit stream according to input, sends task start instruction to one or more from nodes and includes：
Select and for executing the number of nodes of deep learning task from all from nodes being connected with this main controlled node
Suitable from node；
According to deep learning subtask type and all types of subtask quantity, determine that start in each from node appoints
Business；
Send and the task corresponding task start instruction starting in this from node to the from node of each selection.
According to another aspect of the present invention, there is provided a kind of meshed network self-organizing device, wherein, this device is deployed in point
In the from node of cloth cluster, including：
Communication unit, is suitable to the task start instruction of receiving node self-organization of network server transmission；
Unit is chosen in port, is suitable to choose the port numbers for this task；
Described communication unit, the port numbers being further adapted for the host name of this device place from node and choosing return to described
Meshed network hoc service device, so that the described meshed network hoc service device task of being started according to each from node and each save
Host name and port numbers that spot net self-organizing device returns, generate meshed network figure；And be suitable to receive described meshed network
The meshed network figure that hoc service device sends.
Alternatively, unit is chosen in described port, is suitable to the currently unappropriated port numbers from this device place from node
In, randomly select a port number.
Alternatively, described communication unit, is suitable to periodically send acquisition node net to described meshed network hoc service device
The request of network figure, receives the meshed network figure that described meshed network hoc service device returns according to this request；And/or, receive
The meshed network figure that described meshed network hoc service device active issues.
Alternatively, described communication unit, is further adapted for the meshed network sending according to described meshed network hoc service device
Figure, is set up with the meshed network self-organizing device in other one or more from nodes of this meshed network in figure and is connected.
Alternatively, described task start instructs the enabled instruction for deep learning subtask；Described deep learning subtask
Including：Parameter server subtask and/or worker subtask.
According to another aspect of the invention, there is provided a kind of meshed network hoc service device, wherein, this server disposition
On the main controlled node of distributed type assemblies, including：
Communication unit, is suitable to the mission bit stream according to input, to the meshed network self-organizing in one or more from nodes
Device sends task start instruction；Receive host name and the port numbers that each meshed network self-organizing device returns；
Meshed network figure signal generating unit, is suitable to the task according to the startup of each from node and each meshed network self-organizing device returns
The host name returned and port numbers, generate meshed network figure；
Described communication unit, is further adapted for the meshed network being sent to described meshed network figure in one or more from nodes
Self-organizing device.
Alternatively, described communication unit, what the meshed network self-organizing device being suitable on receiving from node sent obtains
Take meshed network figure request when, described meshed network figure is sent to the meshed network self-organizing device in this from node；With/
Or, described meshed network figure is sent to the meshed network in all from nodes that the main controlled node being located with book server is connected
Self-organizing device
Alternatively, described mission bit stream is the mission bit stream of deep learning task；Described mission bit stream includes：For executing
The number of nodes of deep learning task, deep learning subtask type, all types of subtask quantity.
Alternatively, this server also includes：
Scheduling unit, is suitable to select from all from nodes being connected with the main controlled node that book server is located and be used for holding
The suitable from node of the number of nodes of row deep learning task；According to deep learning subtask type and all types of subtask numbers
Amount, determines starting in each from node of task；
Described communication unit, is suitable to send and in this from node to the meshed network self-organizing device in the from node selecting
The task corresponding task start instruction of upper startup.
According to another aspect of the invention, there is provided a kind of meshed network self-organizing system, wherein, this system includes one
Or multiple meshed network self-organizing device as described in any of the above-described and the meshed network self-organizing as described in any of the above-described
Server.
From the foregoing, technical scheme, from node is receiving the task start instruction of main controlled node transmission
Afterwards, actively choose the port numbers for this task, the port numbers of the host name of this from node and selection returned to main controlled node,
So that the host name of the main controlled node task of starting according to each from node and return and port numbers, generate meshed network figure.This skill
The distribution of port numbers is changed to each from node by main controlled node by art scheme, it is to avoid the port numbers of main controlled node distribution with from section
The port numbers of point use clash, and the meshed network figure simultaneously generating can be used between management and the from node of meshed network
Connection establishment, both met use demand, improve simultaneously meshed network structure success rate.
Described above is only the general introduction of technical solution of the present invention, in order to better understand the technological means of the present invention,
And can be practiced according to the content of specification, and in order to allow the above and other objects of the present invention, feature and advantage can
Become apparent, below especially exemplified by the specific embodiment of the present invention.
Brief description
By reading the detailed description of hereafter preferred embodiment, various other advantages and benefit are common for this area
Technical staff will be clear from understanding.Accompanying drawing is only used for illustrating the purpose of preferred embodiment, and is not considered as to the present invention
Restriction.And in whole accompanying drawing, it is denoted by the same reference numerals identical part.In the accompanying drawings：
Fig. 1 shows a kind of schematic flow sheet of meshed network self-organizing method according to an embodiment of the invention；
Fig. 2 shows the schematic flow sheet of another kind meshed network self-organizing method according to an embodiment of the invention；
Fig. 3 shows a kind of structural representation of meshed network self-organizing device according to an embodiment of the invention；
Fig. 4 shows a kind of structural representation of meshed network hoc service device according to an embodiment of the invention；
Fig. 5 shows a kind of structural representation of meshed network self-organizing system according to an embodiment of the invention.
Specific embodiment
It is more fully described the exemplary embodiment of the disclosure below with reference to accompanying drawings.Although showing the disclosure in accompanying drawing
Exemplary embodiment it being understood, however, that may be realized in various forms the disclosure and should not be by embodiments set forth here
Limited.On the contrary, these embodiments are provided to be able to be best understood from the disclosure, and can be by the scope of the present disclosure
Complete conveys to those skilled in the art.
Fig. 1 shows a kind of schematic flow sheet of meshed network self-organizing method according to an embodiment of the invention, such as
Shown in Fig. 1, the method includes：
Step S110, receives the task start instruction that main controlled node sends.
Step S120, chooses the port numbers for this task.
Step S130, the port numbers of the host name of this from node and selection are returned to main controlled node, so that main controlled node
Being started according to each from node of task and the host name of return and port numbers, generate meshed network figure.
Step S140, receives the meshed network figure that main controlled node sends.
It can be seen that, the method shown in Fig. 1, from node, after the task start instruction receiving main controlled node transmission, is actively selected
Take the port numbers in this task, the port numbers of the host name of this from node and selection are returned to main controlled node, so that master control
Task and the host name of return and port numbers that node starts according to each from node, generate meshed network figure.This technical scheme will
The distribution of port numbers is changed to each from node by main controlled node, it is to avoid the port numbers of main controlled node distribution are used with from node
Port numbers clash, the connection that the meshed network figure simultaneously generating can be used between the management of meshed network and from node is built
Vertical, both meet use demand, improve the success rate of meshed network structure simultaneously.
In one embodiment of the invention, in the method shown in Fig. 1, choose the port numbers for this task and include：From
This from node currently, in unappropriated port numbers, randomly selects a port number.
Port numbers in from node are usually 65535 circulations, choose at random, efficiency high from unappropriated port numbers,
Port collision will not be produced simultaneously.
In one embodiment of the invention, in the method shown in Fig. 1, receive the meshed network figure bag that main controlled node sends
Include：Periodically send, to main controlled node, the request obtaining meshed network figure, receive the node net that main controlled node returns according to this request
Network figure；And/or, receive the meshed network figure that main controlled node active issues.
For example, sent, to main controlled node, the request obtaining meshed network figure every 10 minutes, or by main controlled node in section
After spot net figure changes, the meshed network figure after updating is issued to the from node of correlation.
The major reason that each from node obtains meshed network figure is that under many circumstances, what from node was run enters
Journey needs to be communicated with the process in other from nodes.Therefore in one embodiment of the invention, the method shown in Fig. 1 is also
Including：The meshed network figure being sent according to main controlled node, is set up with other one or more from nodes of this meshed network in figure
Connect.
In one embodiment of the invention, in the method shown in Fig. 1, task start instructs as deep learning subtask
Enabled instruction；Deep learning subtask includes：Parameter server subtask and/or worker subtask.In this instance,
Parameter server, as parameter server, needs to receive the calculated parameter that worker subtask is submitted to.
Fig. 2 shows the schematic flow sheet of another kind meshed network self-organizing method according to an embodiment of the invention,
As shown in Fig. 2 the method includes：
Step S210, according to the mission bit stream of input, sends task start instruction to one or more from nodes.
Step S220, receives host name and the port numbers that each from node returns.
Step S230, being started according to each from node of task and the host name of return and port numbers, generate meshed network figure.
Step S240, meshed network figure is sent to one or more from nodes.
In one embodiment of the invention, in the method shown in Fig. 2, by meshed network figure be sent to one or more from
Node includes：When receiving the request of acquisition meshed network figure of from node transmission, meshed network figure is sent to this from section
Point；And/or, meshed network figure is sent to all from nodes being connected with this main controlled node.
Provide the distribution method of two kinds of meshed network figures in the present embodiment, can be used in combination, but do not represent right
The restriction of distribution method is it is also possible to when meshed network figure occurs change, the meshed network figure after updating is only sent to this
Update related node.For example, task A has increased two execution nodes, node 13 and node 14 newly, and task A former execution node is
Node 1 and node 2, then only need to for the meshed network figure after updating to be sent to node 1,3,13 and 14.Certainly, meshed network
Figure can also generate corresponding component according to each task, so only needs to when issuing meshed network figure distribute them to this figure
Middle related node.
In one embodiment of the invention, in the method shown in Fig. 2, mission bit stream is the task letter of deep learning task
Breath；Mission bit stream includes：For executing the number of nodes of deep learning task, deep learning subtask type, all types of sons
Task quantity.
Deep learning task is to carry out the submission of calculating task in graph form, and these tasks upon execution can be further
It is divided into multiple operations, each operation includes one or more subtasks.Subtask type includes one or more of following：
Parameter server subtask, worker subtask.
For example, TensorFlow is exactly a deep learning storehouse increased income.Tensor (tensor) means N-dimensional array,
Flow (stream) means the calculating based on DFD, and TensorFlow is that tensor flow to other end calculating from one end of image
Process.This deep learning storehouse can be integrated with Spark big data Computational frame, will a TensorFlow task conduct
One Spark task is submitted to, that is, deep learning task alleged above.Deep learning mission bit stream can also include
One or more of following：The calculating figure of execution deep learning；The deep learning storehouse that execution deep learning task need to be called connects
Mouthful；Data address for deep learning task；The preservation address of implementing result data.
In one embodiment of the invention, in said method, according to the mission bit stream of input, to one or more from section
Point sends task start instruction and includes：Select and for executing deep learning from all from nodes being connected with this main controlled node
The suitable from node of the number of nodes of task；According to deep learning subtask type and all types of subtask quantity, determine
Starting in each from node of task；Send to the from node of each selection and open with the corresponding task of task starting in this from node
Dynamic instruction.
Taking a deep learning task as a example, if according to the mission bit stream of this deep learning task, need to call depth
Learning database starts 2 parameter server subtasks and 2 worker subtasks, and this four subtasks are respectively four
Execute in individual from node, then just first determine the subtask of execution in each task, then send startup phase to each from node
The instruction of the subtask answered.
It is mentioned above, deep learning storehouse can be integrated with Spark big data Computational frame, and that is, distributed type assemblies are permissible
For Spark cluster.Spark cluster can also carry out scheduling, job management and the resource management of task by Yarn.Yarn is permissible
Provide the user front end page for the submission of task, therefore in one embodiment of the invention, the deep learning of submission is appointed
Business can be inputted by front end page.After task start, the front end page that user can also provide according to Yarn, in real time
Check the treatment situation of task, task is carried out kill etc. with operation.Because Spark cluster can also carry out task by Yarn
Scheduling, job management and resource management, therefore in above-described embodiment, are selected from all from nodes being connected with this main controlled node
Can also be obtained currently by sending request to Yarn with for executing the suitable from node of the number of nodes of deep learning task
More idle node is executing deep learning task.I.e.：Send for executing this depth to the node scheduling device of distributed type assemblies
The number of nodes of degree learning tasks, and the information of multiple nodes of receiving node scheduler return.
The example of the meshed network figure that for a deep learning task generate has been illustrated below：
{PS:[node1:8080,node2:8080]worker:[node3:9090,node4:9090]}
This means parameter server subtask is started on 8080 ports of node 1, the 8080 of node 2
Parameter server subtask is started on port；Worker subtask is started on 9090 ports of node 3, in section
Worker subtask is started on 9090 ports of point 4.Next need actively meshed network figure to be handed down to these from nodes,
Or obtain request according to being sent by each from node from favourite network list, meshed network figure is handed down to these nodes.Example
As, on 9090 ports of node 3 start worker subtask can respectively with 8080 ports of node 1 on start
Parameter server subtask and the parameter server subtask starting on 8080 ports of node 2 are set up
Connect.
These can start a Driver process by Spark, start one simultaneously after deep learning task is submitted to
Scheduler dispatches process, is realized structure, management and the distribution of meshed network figure by this process.
Specifically, obtain the data for this deep learning task from the file system of distributed type assemblies to include：According to
For the data address of deep learning task, the data structure of this deep learning task in the file system of distributed type assemblies, will be used for
Build as elasticity distribution formula data set RDD object；The data-pushing being used for this deep learning task obtaining is appointed to son accordingly
Carry out in business executing including：RDD object is pushed to each node respectively, by each node by RDD Object Push in this node
On the subtask starting.
, its data storage is in HDFS (Hadoop Distributed File taking Spark distributed type assemblies as a example
System, Hadoop distributed file system) on.In peration data, it is configured to accordingly a RDD
(resilientdistributed dataset, elasticity distribution formula data set) object.RDD object can be multiplexed, if depth
Data used by learning tasks has been built as RDD object, then naturally avoid the need for executing this step.Using these data
When, pushed it in the from node that each task is located by pipeline (pipe), by each node by RDD Object Push to this from
On the subtask starting in node.As a example deep learning task in above example comprises two worker subtasks, need RDD
A part for object is pushed on node 3, and another part is pushed on node 4, it is achieved thereby that distributed treatment deep learning
Task.
Fig. 3 shows a kind of structural representation of meshed network self-organizing device according to an embodiment of the invention, should
Device can be deployed in the from node of distributed type assemblies.As shown in figure 3, meshed network self-organizing device 300 includes：
Communication unit 310, is suitable to the task start instruction of receiving node self-organization of network server transmission.
Unit 320 is chosen in port, is suitable to choose the port numbers for this task.
Communication unit 310, the port numbers being further adapted for the host name of this device place from node and choosing return to node
Self-organization of network server so that the meshed network hoc service device task of being started according to each from node and each meshed network from
Host name and port numbers that tissue device returns, generate meshed network figure；And it is suitable to receiving node self-organization of network server
The meshed network figure sending.
It can be seen that, the device shown in Fig. 3, cooperating by each unit, from node is receiving main controlled node transmission
After task start instruction, actively choose the port numbers for this task, the port numbers of the host name of this from node and selection are returned
Back to main controlled node, so that the host name of the main controlled node task of starting according to each from node and return and port numbers, generate section
Spot net figure.The distribution of port numbers is changed to each from node by main controlled node by this technical scheme, it is to avoid main controlled node distribution
Port numbers clash with the port numbers of from node use, the meshed network figure simultaneously generating can be used for meshed network
Connection establishment between management and from node, had both met use demand, improve the success rate of meshed network structure simultaneously.
In one embodiment of the invention, in the device shown in Fig. 3, unit 320 is chosen in port, is suitable to from this device institute
In the currently unappropriated port numbers of from node, randomly select a port number.
In one embodiment of the invention, in the device shown in Fig. 3, communication unit 310, it is suitable to periodically to meshed network
Hoc service device sends the request obtaining meshed network figure, and receiving node self-organization of network server returns according to this request
Meshed network figure；And/or, the meshed network figure that receiving node self-organization of network server active issues.
In one embodiment of the invention, in the device shown in Fig. 3, communication unit 310, it is further adapted for according to meshed network
The meshed network figure that hoc service device sends, with the node net in other one or more from nodes of this meshed network in figure
Network self-organizing device 300 is set up and is connected.
In one embodiment of the invention, in the device shown in Fig. 3, task start instructs as deep learning subtask
Enabled instruction；Deep learning subtask includes：Parameter server subtask and/or worker subtask.
Fig. 4 shows a kind of structural representation of meshed network hoc service device according to an embodiment of the invention,
This server can be deployed on the main controlled node of distributed type assemblies.As shown in figure 4, meshed network hoc service device 400 wraps
Include：
Communication unit 410, is suitable to the mission bit stream according to input, to the meshed network in one or more from nodes from group
Knit device and send task start instruction；Receive host name and the port numbers that each meshed network self-organizing device returns；
Meshed network figure signal generating unit 420, is suitable to task and each meshed network self-organizing dress starting according to each from node
Put host name and the port numbers of return, generate meshed network figure；
Communication unit 410, is further adapted for for meshed network figure being sent to meshed network in one or more from nodes from group
Knit device.
In one embodiment of the invention, in the server shown in Fig. 4, communication unit 410, it is suitable to receiving from section
During the request of acquisition meshed network figure that the meshed network self-organizing device on point sends, meshed network figure is sent to this from section
Meshed network self-organizing device on point；And/or, meshed network figure is sent to the main controlled node being located with book server and is connected
All from nodes on meshed network self-organizing device
In one embodiment of the invention, in the server shown in Fig. 4, mission bit stream is the task of deep learning task
Information；Mission bit stream includes：For executing the number of nodes of deep learning task, deep learning subtask type, all types of
Subtask quantity.
In one embodiment of the invention, also include in the server shown in Fig. 4：Scheduling unit 430, be suitable to from this
Select and the number of nodes phase for executing deep learning task in all from nodes that the main controlled node that server is located connects
When from node；According to deep learning subtask type and all types of subtask quantity, determine startup in each from node
Task；Communication unit 310, is suitable to send to the meshed network self-organizing device in the from node selecting and opens with this from node
The corresponding task start of dynamic task instructs.
It should be noted that the specific embodiment of above-mentioned each device server embodiment is implemented with aforementioned corresponding method
The specific embodiment of example is similar, will not be described here.Slightly different, each from node not only can be deployed with node
Self-organization of network device, can be with the performs device of deployment task.Meshed network not only can be deployed with main controlled node from group
Knit server, can be with the control server of deployment task.Certainly, these servers can be integrated as one each via function
Individual server realizing, can also be integrated by function and be used as a device and realize by each device in same from node.
Fig. 5 shows a kind of structural representation of meshed network self-organizing system according to an embodiment of the invention, such as
Shown in Fig. 5, meshed network self-organizing system 500 includes the meshed network self-organizing in one or more such as above-mentioned any embodiment
Meshed network hoc service device 400 in device 300 and such as above-mentioned any embodiment.
In sum, technical scheme, from node receive main controlled node transmission task start instruction after,
Actively choose the port numbers for this task, the port numbers of the host name of this from node and selection are returned to main controlled node, with
Make host name and the port numbers of task that main controlled node starts according to each from node and return, generate meshed network figure.This technology
The distribution of port numbers is changed to each from node by main controlled node by scheme, it is to avoid the port numbers of main controlled node distribution and from node
The port numbers of use clash, and the meshed network figure simultaneously generating can be used between management and the from node of meshed network
Connection establishment, had both met use demand, improve the success rate of meshed network structure simultaneously.
It should be noted that：
Algorithm and display be not inherently related to any certain computer, virtual bench or miscellaneous equipment provided herein.
Various fexible units can also be used together with based on teaching in this.As described above, construct required by this kind of device
Structure be obvious.Additionally, the present invention is also not for any certain programmed language.It is understood that, it is possible to use various
Programming language realizes the content of invention described herein, and the description above language-specific done is to disclose this
Bright preferred forms.
In specification mentioned herein, illustrate a large amount of details.It is to be appreciated, however, that the enforcement of the present invention
Example can be put into practice in the case of not having these details.In some instances, known method, structure are not been shown in detail
And technology, so as not to obscure the understanding of this description.
Similarly it will be appreciated that in order to simplify the disclosure and help understand one or more of each inventive aspect,
Above in the description to the exemplary embodiment of the present invention, each feature of the present invention is grouped together into single enforcement sometimes
In example, figure or descriptions thereof.However, the method for the disclosure should be construed to reflect following intention：I.e. required guarantor
The application claims of shield more features than the feature being expressly recited in each claim.More precisely, it is such as following
Claims reflected as, inventive aspect is all features less than single embodiment disclosed above.Therefore,
The claims following specific embodiment are thus expressly incorporated in this specific embodiment, wherein each claim itself
All as the separate embodiments of the present invention.
Those skilled in the art are appreciated that and the module in the equipment in embodiment can be carried out adaptively
Change and they are arranged in one or more equipment different from this embodiment.Can be the module in embodiment or list
Unit or assembly be combined into a module or unit or assembly, and can be divided in addition multiple submodule or subelement or
Sub-component.In addition to such feature and/or at least some of process or unit exclude each other, can adopt any
Combination is to all features disclosed in this specification (including adjoint claim, summary and accompanying drawing) and so disclosed
Where method or all processes of equipment or unit are combined.Unless expressly stated otherwise, this specification (includes adjoint power
Profit requires, summary and accompanying drawing) disclosed in each feature can carry out generation by the alternative features providing identical, equivalent or similar purpose
Replace.
Although additionally, it will be appreciated by those of skill in the art that some embodiments described herein include other embodiments
In included some features rather than further feature, but the combination of the feature of different embodiment means to be in the present invention's
Within the scope of and form different embodiments.For example, in the following claims, embodiment required for protection appoint
One of meaning can in any combination mode using.
The all parts embodiment of the present invention can be realized with hardware, or to run on one or more processor
Software module realize, or with combinations thereof realize.It will be understood by those of skill in the art that can use in practice
Microprocessor or digital signal processor (DSP) are realizing meshed network self-organizing device according to embodiments of the present invention, clothes
The some or all functions of some or all parts in business device and system.The present invention is also implemented as executing this
In described some or all equipment of method or program of device (for example, computer program and computer program
Product).Such program realizing the present invention can store on a computer-readable medium, or can have one or many
The form of individual signal.Such signal can be downloaded from internet website and obtain, or provides on carrier signal, or with
Any other form provides.
It should be noted that above-described embodiment the present invention will be described rather than limits the invention, and ability
Field technique personnel can design alternative embodiment without departing from the scope of the appended claims.In the claims,
Any reference symbol between bracket should not be configured to limitations on claims.Word "comprising" does not exclude the presence of not
Element listed in the claims or step.Word "a" or "an" before element does not exclude the presence of multiple such
Element.The present invention can come real by means of the hardware including some different elements and by means of properly programmed computer
Existing.If in the unit claim listing equipment for drying, several in these devices can be by same hardware branch
To embody.The use of word first, second, and third does not indicate that any order.These words can be explained and run after fame
Claim.
Embodiment of the invention discloses that A1, a kind of meshed network self-organizing method, wherein, the method includes：
Receive the task start instruction that main controlled node sends；
Choose the port numbers for this task；
The port numbers of the host name of this from node and selection are returned to described main controlled node, so that described main controlled node root
Starting according to each from node of task and the host name of return and port numbers, generate meshed network figure；
Receive the meshed network figure that described main controlled node sends.
A2, the method as described in A1, wherein, the described port numbers chosen for this task include：
From this from node currently unappropriated port numbers, randomly select a port number.
A3, the method as described in A1, wherein, the described meshed network figure receiving described main controlled node transmission includes：
Periodically send, to described main controlled node, the request obtaining meshed network figure, receive described main controlled node according to this request
The meshed network figure returning；
And/or,
Receive the meshed network figure that described main controlled node active issues.
A4, the method as described in A1, wherein, the method also includes：
According to described main controlled node send meshed network figure, with this meshed network in figure other one or more from section
Point is set up and is connected.
A5, the method as described in A1, wherein,
Described task start instructs the enabled instruction for deep learning subtask；Described deep learning subtask includes：
Parameter server subtask and/or worker subtask.
Embodiments of the invention also disclose B6, a kind of meshed network self-organizing method, and wherein, the method includes：
According to the mission bit stream of input, send task start instruction to one or more from nodes；
Receive host name and the port numbers that each from node returns；
Being started according to each from node of task and the host name of return and port numbers, generate meshed network figure；
Described meshed network figure is sent to one or more from nodes.
Described meshed network figure wherein, described is sent to one or more from node bags by B7, the method as described in B6
Include：
When receive from node transmission acquisition meshed network figure request when, by described meshed network figure be sent to this from
Node；
And/or,
Described meshed network figure is sent to all from nodes being connected with this main controlled node.
B8, the method as described in B6, wherein, described mission bit stream is the mission bit stream of deep learning task；Described task
Information includes：For executing the number of nodes of deep learning task, deep learning subtask type, all types of subtask numbers
Amount.
B9, the method as described in B8, wherein, the described mission bit stream according to input, send to one or more from nodes
Task start instruction includes：
Select and for executing the number of nodes of deep learning task from all from nodes being connected with this main controlled node
Suitable from node；
According to deep learning subtask type and all types of subtask quantity, determine that start in each from node appoints
Business；
Send and the task corresponding task start instruction starting in this from node to the from node of each selection.
Embodiments of the invention also disclose C10, a kind of meshed network self-organizing device, and wherein, this device is deployed in point
In the from node of cloth cluster, including：
Communication unit, is suitable to the task start instruction of receiving node self-organization of network server transmission；
Unit is chosen in port, is suitable to choose the port numbers for this task；
Described communication unit, the port numbers being further adapted for the host name of this device place from node and choosing return to described
Meshed network hoc service device, so that the described meshed network hoc service device task of being started according to each from node and each save
Host name and port numbers that spot net self-organizing device returns, generate meshed network figure；And be suitable to receive described meshed network
The meshed network figure that hoc service device sends.
C111, the device as described in C110, wherein,
Unit is chosen in described port, is suitable to from the currently unappropriated port numbers of this device place from node, at random
Choose a port number.
C112, the device as described in C110, wherein,
Described communication unit, is suitable to periodically send asking of acquisition meshed network figure to described meshed network hoc service device
Ask, receive the meshed network figure that described meshed network hoc service device returns according to this request；And/or, receive described node
The meshed network figure that self-organization of network server active issues.
C113, the device as described in C110, wherein,
Described communication unit, is further adapted for the meshed network figure sending according to described meshed network hoc service device, with this
Meshed network self-organizing device in other one or more from nodes of meshed network in figure is set up and is connected.
C114, the device as described in C110, wherein,
Described task start instructs the enabled instruction for deep learning subtask；Described deep learning subtask includes：
Parameter server subtask and/or worker subtask.
Embodiments of the invention also disclose D15, a kind of meshed network hoc service device, wherein, this server disposition
On the main controlled node of distributed type assemblies, including：
Communication unit, is suitable to the mission bit stream according to input, to the meshed network self-organizing in one or more from nodes
Device sends task start instruction；Receive host name and the port numbers that each meshed network self-organizing device returns；
Meshed network figure signal generating unit, is suitable to the task according to the startup of each from node and each meshed network self-organizing device returns
The host name returned and port numbers, generate meshed network figure；
Described communication unit, is further adapted for the meshed network being sent to described meshed network figure in one or more from nodes
Self-organizing device.
D16, the server as described in D15, wherein,
Described communication unit, the acquisition node net that the meshed network self-organizing device being suitable on receiving from node sends
During the request of network figure, described meshed network figure is sent to the meshed network self-organizing device in this from node；And/or, by institute
State the meshed network self-organizing that meshed network figure is sent in all from nodes that the main controlled node being located with book server is connected
Device
D17, the server as described in D15, wherein, described mission bit stream is the mission bit stream of deep learning task；Described
Mission bit stream includes：For executing the number of nodes of deep learning task, deep learning subtask type, all types of subtasks
Quantity.
D18, the server as described in D17, wherein, this server also includes：
Scheduling unit, is suitable to select from all from nodes being connected with the main controlled node that book server is located and be used for holding
The suitable from node of the number of nodes of row deep learning task；According to deep learning subtask type and all types of subtask numbers
Amount, determines starting in each from node of task；
Described communication unit, is suitable to send and in this from node to the meshed network self-organizing device in the from node selecting
The task corresponding task start instruction of upper startup.
Embodiments of the invention also disclose E19, a kind of meshed network self-organizing system, and wherein, this system includes one
Or multiple meshed network self-organizing device as any one of C10-C14 and as any one of claim D15-D18 institute
The meshed network hoc service device stated.
Claims (10)
1. a kind of meshed network self-organizing method, wherein, the method includes：
Receive the task start instruction that main controlled node sends；
Choose the port numbers for this task；
The port numbers of the host name of this from node and selection are returned to described main controlled node, so that described main controlled node is according to each
Task and the host name of return and port numbers that from node starts, generate meshed network figure；
Receive the meshed network figure that described main controlled node sends.
2. the method for claim 1, wherein the described port numbers chosen for this task include：
From this from node currently unappropriated port numbers, randomly select a port number.
3. the method for claim 1, wherein the described meshed network figure receiving described main controlled node transmission includes：
Periodically send, to described main controlled node, the request obtaining meshed network figure, receive described main controlled node and returned according to this request
Meshed network figure；
And/or,
Receive the meshed network figure that described main controlled node active issues.
4. a kind of meshed network self-organizing method, wherein, the method includes：
According to the mission bit stream of input, send task start instruction to one or more from nodes；
Receive host name and the port numbers that each from node returns；
Being started according to each from node of task and the host name of return and port numbers, generate meshed network figure；
Described meshed network figure is sent to one or more from nodes.
5. described meshed network figure wherein, described is sent to one or more from node bags by method as claimed in claim 4
Include：
When receiving the request of acquisition meshed network figure of from node transmission, described meshed network figure is sent to this from section
Point；
And/or,
Described meshed network figure is sent to all from nodes being connected with this main controlled node.
6. a kind of meshed network self-organizing device, wherein, this device is deployed in the from node of distributed type assemblies, including：
Communication unit, is suitable to the task start instruction of receiving node self-organization of network server transmission；
Unit is chosen in port, is suitable to choose the port numbers for this task；
Described communication unit, the port numbers being further adapted for the host name of this device place from node and choosing return to described node
Self-organization of network server, so that the described meshed network hoc service device task of being started according to each from node and each node net
Host name and port numbers that network self-organizing device returns, generate meshed network figure；And be suitable to receive described meshed network from group
Knit the meshed network figure of server transmission.
7. device as claimed in claim 6, wherein,
Unit is chosen in described port, is suitable to from the currently unappropriated port numbers of this device place from node, randomly selects
A port number.
8. a kind of meshed network hoc service device, wherein, this server disposition, on the main controlled node of distributed type assemblies, wraps
Include：
Communication unit, is suitable to the mission bit stream according to input, to the meshed network self-organizing device in one or more from nodes
Send task start instruction；Receive host name and the port numbers that each meshed network self-organizing device returns；
Meshed network figure signal generating unit, be suitable to according to each from node start task and each meshed network self-organizing device return
Host name and port numbers, generate meshed network figure；
Described communication unit, is further adapted for for described meshed network figure being sent to meshed network in one or more from nodes from group
Knit device.
9. server as claimed in claim 8, wherein,
Described communication unit, the acquisition meshed network figure that the meshed network self-organizing device being suitable on receiving from node sends
Request when, described meshed network figure is sent to the meshed network self-organizing device in this from node；And/or, by described section
Spot net figure is sent to the meshed network self-organizing device in all from nodes that the main controlled node being located with book server is connected.
10. a kind of meshed network self-organizing system, wherein, this system includes one or more such as any one of claims 6-7
Described meshed network self-organizing device and the meshed network hoc service device as any one of claim 8-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610963725.XA CN106411605B (en) | 2016-10-28 | 2016-10-28 | A kind of meshed network self-organizing method, device, server and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610963725.XA CN106411605B (en) | 2016-10-28 | 2016-10-28 | A kind of meshed network self-organizing method, device, server and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106411605A true CN106411605A (en) | 2017-02-15 |
| CN106411605B CN106411605B (en) | 2019-05-31 |
Family
ID=58014284
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610963725.XA Active CN106411605B (en) | 2016-10-28 | 2016-10-28 | A kind of meshed network self-organizing method, device, server and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106411605B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109614232A (en) * | 2018-12-07 | 2019-04-12 | 网易（杭州）网络有限公司 | Task processing method, device, storage medium and electronic device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060015643A1 (en) * | 2004-01-23 | 2006-01-19 | Fredrik Orava | Method of sending information through a tree and ring topology of a network system |
| CN101888331A (en) * | 2009-05-13 | 2010-11-17 | 阿瓦亚公司 | Be used to provide the method and apparatus of the quick rerouting of unicast packet |
| CN104158747A (en) * | 2013-05-14 | 2014-11-19 | 中兴通讯股份有限公司 | Network topology discovery method and system |
| CN105391580A (en) * | 2015-11-27 | 2016-03-09 | 上海卫星工程研究所 | Network model description method applicable to SpW/SpF |
| CN105721318A (en) * | 2016-02-29 | 2016-06-29 | 华为技术有限公司 | Method and device for discovering network topology in software defined network SDN (Software Defined Network) |
-
2016
- 2016-10-28 CN CN201610963725.XA patent/CN106411605B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060015643A1 (en) * | 2004-01-23 | 2006-01-19 | Fredrik Orava | Method of sending information through a tree and ring topology of a network system |
| CN101888331A (en) * | 2009-05-13 | 2010-11-17 | 阿瓦亚公司 | Be used to provide the method and apparatus of the quick rerouting of unicast packet |
| CN104158747A (en) * | 2013-05-14 | 2014-11-19 | 中兴通讯股份有限公司 | Network topology discovery method and system |
| CN105391580A (en) * | 2015-11-27 | 2016-03-09 | 上海卫星工程研究所 | Network model description method applicable to SpW/SpF |
| CN105721318A (en) * | 2016-02-29 | 2016-06-29 | 华为技术有限公司 | Method and device for discovering network topology in software defined network SDN (Software Defined Network) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109614232A (en) * | 2018-12-07 | 2019-04-12 | 网易（杭州）网络有限公司 | Task processing method, device, storage medium and electronic device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106411605B (en) | 2019-05-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108958920B (en) | Distributed task scheduling method and system | |
| Yang et al. | Delay-aware virtual network function placement and routing in edge clouds | |
| EP2710470B1 (en) | Extensible centralized dynamic resource distribution in a clustered data grid | |
| CN105100259B (en) | A kind of distributed timing task executing method and system | |
| Houidi et al. | A distributed virtual network mapping algorithm | |
| KR102163402B1 (en) | System for executing distributed deep learning using multi node and multi graphics processing unit and method thereof | |
| US20090070773A1 (en) | Method for efficient thread usage for hierarchically structured tasks | |
| EP3596600B1 (en) | Systems and methods for compute node management protocols | |
| US20050228855A1 (en) | Acquisition system for distributed computing resources | |
| CN106529682A (en) | Method and apparatus for processing deep learning task in big-data cluster | |
| CN113382077B (en) | Micro-service scheduling method, micro-service scheduling device, computer equipment and storage medium | |
| CN109471896A (en) | Data source information dynamic altering method and device | |
| CN109656688B (en) | Method, system and server for realizing distributed business rules | |
| CN110022230A (en) | The parallel dispositions method of service chaining and device based on deeply study | |
| CN109101334A (en) | A kind of micro services concurrency control method towards Zuul gateway | |
| US20230082372A1 (en) | Enhanced self-assembling and self-configuring microservices | |
| CN110601994A (en) | Load balancing method for micro-service chain perception in cloud environment | |
| CN105550028B (en) | A kind of multitask timing execution method and system based on buffer lock | |
| CN106411605A (en) | Node network self-organizing method, apparatus, server and system | |
| CN107645396B (en) | Cluster capacity expansion method and device | |
| Abreu et al. | Resilient service chains through smart replication | |
| CN115567425B (en) | Internet distributed active detection method and system | |
| Fan et al. | Associated task scheduling based on dynamic finish time prediction for cloud computing | |
| US20170147401A1 (en) | Api and user interface for mapreduce jobs | |
| Santana-Santana et al. | Load balancing algorithms with partial information management for the dlml library |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220725 Address after: Room 801, 8th floor, No. 104, floors 1-19, building 2, yard 6, Jiuxianqiao Road, Chaoyang District, Beijing 100015 Patentee after: BEIJING QIHOO TECHNOLOGY Co.,Ltd. Address before: 100088 room 112, block D, 28 new street, new street, Xicheng District, Beijing (Desheng Park) Patentee before: BEIJING QIHOO TECHNOLOGY Co.,Ltd. Patentee before: Qizhi software (Beijing) Co.,Ltd. |
|
| TR01 | Transfer of patent right |